Marginal Jobs and Job Surplus: Evidence from Separations and Unemployment Insurance

We study the role of marginal jobs in employment adjustment, in three steps. First, we provide evidence on job destruction in response to reductions in job surplus from improved worker outside options. Our design exploits a sharp quasi-experimental increase in unemployment benefits for older workers in Austria. The treatment effect is larger for workers with larger outside option increases, proxied for with their ex-ante risk of exhausting the pre-reform maximum benefit duration when unemployed. Second, we isolate and characterize the marginal matches driving this separation response, extending complier analysis to difference-in-difference settings. We find that marginal jobs originate from blue- collar occupations in industries with a high incidence of sickness and disability among older workers. Compared to surviving jobs, marginal jobs had lower earnings and lower worker fixed effects and were more prevalent in shrinking industries and firms. Taken together, our findings indicate that increasing workers’ outside options destroys low-surplus jobs. Third, one direct implication is that outside options shift the composition of surviving jobs towards higher-surplus jobs. To test this prediction, we exploit the abolition of the reform to show that the formerly-treated cohorts indeed exhibit lower extensive-margin aggregate elasticities to subsequent labor demand shocks – due to the missing mass of marginal matches the reform had previously destroyed.